CT truncation artifact removal using water-equivalent thicknesses derived from truncated projection data.

Area of Science:

• Medical Imaging
• Image Reconstruction
• Computational Imaging

Background:

• Limited field-of-view (FOV) in computed tomography (CT) imaging leads to projection truncation.
• Truncation artifacts, such as central cupping and bright external rings, degrade image quality.
• Increasing FOV often requires advanced hardware and full angular scanning.

Purpose of the Study:

• To develop a novel method for completing truncated CT projections.
• To utilize water-equivalent thickness (WET) as auxiliary information to correct for truncation artifacts.
• To improve quantitative image quality in CT reconstructions.

Main Methods:

• Estimating patient thickness along projection rays using WET, unaffected by truncation.
• Parameterizing points along rays intersecting the object boundary, separated by WET.
• Approximating patient outline as an ellipse and simultaneously estimating ellipse parameters and point positions using deterministic optimization.
• Completing truncated projections with the optimal ellipse for subsequent image reconstruction.

Main Results:

• The root-mean-square (RMS) error in reconstructions was reduced from 20.4% to 1.0% for a severely truncated abdominal CT dataset.
• The novel algorithm demonstrated a significant reduction in reconstruction error compared to empirical extrapolation methods.
• Quantitative image quality improvements were substantial despite the assumption of an elliptical patient cross-section.

Conclusions:

• The proposed method effectively completes truncated CT projections, leading to significant improvements in image quality.
• The algorithm is robust, with pelvic bone presence minimally biasing ellipse position.
• The method is computationally efficient and suitable for clinical integration.